I've been wondering what this Tak function was invented for - I imagined it was something like Ackermann(*) which demonstrates some mathematical principle. But no - it was invented to benchmark Lisp!I've been working my way through an article on the tak function in various languages and have come across what seems to be an oddity with LISP.
From a 1979 article by John McCarthy:
AN INTERESTING LISP FUNCTION
Ikuo Takeuchi (1978) of the Electrical Communication Laboratory of Nippon Telephone and Telegraph Co. (Japan's Bell Labs) devised a recursive function program for comparing the speeds of LISP systems. It can be made to run a long time without generating large numbers or using much stack.
Edit: aha, but in this book we're told the TAK function is a mis-remembered version of the original Takeuchi benchmark!
main text:
footnote:The TAK benchmark is a variant of the Takeuchi function that Ikuo Takeuchi of Japan used as a simple benchmark.
(*) Wikipedia says "one of the simplest and earliest-discovered examples of a total computable function that is not primitive recursive"Historical note: When the Computer Science Department at Stanford University obtained the first two or three Xerox Dolphins, John McCarthy asked me to do a simple benchmark test with him. We sat down, and he tried to remember the Takeuchi function, which had had wide circulation. Because it was simple and because there were many results for it in the literature, he felt that it would be a good initial test. Of course, John misremembered the function. But we did not realize it until I had gathered a great many numbers for it. Alas, we are stuck with this variant on Takeuchi.
Statistics: Posted by BigEd — Thu Jan 16, 2025 8:56 pm